The Dispersion of Echinococcus granulosus in the Intestine ...€¦ · Adults of Echinococcus granulosus (Cestoda: Taeniidae) develop from ingested protoscoleces within the small

J. Parasitol., 75(4), 1989, p. 562-570

? American Society of Parasitologists 1989

THE DISPERSION OF ECHINOCOCCUS GRANULOSUS IN THE INTESTINE OF DOGS

A. J. Lymbery, R. P. Hobbs, and R. C. A. Thompson Division of Veterinary Biology, School of Veterinary Studies, Murdoch University, Murdoch, 6150, Western Australia

ABSTRACT: We studied the dispersion of adult Echinococcus granulosus in the intestine of experimentally infected dogs at 2 scales of habitat use. On a coarse scale, worms were found most frequently in the anterior third of the small intestine. On a fine scale, clumps or aggregations, typically of 4-5 worms in an area of 12 mm2, occurred throughout the anterior two-thirds of the intestine. The most likely proximate cause of aggregative behavior is attraction between individual worms. There are at least 2 equally plausible ultimate causes of the behavior: to enhance cross-fertilization and to improve the quality of te environment. Restriction of worms to the anterior small intestine may be a consequence of aggregative behavior on a finer scale or a response to different proximate and ultimate factors.

All species of parasites are restricted, not only to particular host species, but also to specific sites on or in the host. Site specificity is usually pre- sumed to result from active site selection by the parasite (Holmes, 1973). Identification of the causes of site selection requires a distinction be- tween proximate and ultimate explanations. The proximate causes are those factors that act as the immediate stimulus for the behavior. These usu- ally are presumed to be specific physical or chem- ical cues, although there is little supporting evi- dence (Mettrick and Podesta, 1974; Sukhdeo and Mettrick, 1987). The ultimate causes of site se- lection are those factors responsible for the evo- lution of the behavior. Site specificity implies restriction along 1 or more dimensions of the niche of a species and there has been much spec- ulation on the function of niche restriction in parasites. Price (1984) argued that species of par- asites are highly specialized to exploit precise microenvironments and that niche restriction is therefore a consequence of choosing an optimal habitat. Holmes (1973) presented evidence, largely from cestodes and nematodes, that inter- specific competition is responsible for niche re- striction. Rohde (1979) and Rohde and Hobbs (1986) argued, primarily from data on mono- geneans, that niche restriction functions to en- hance intraspecific contact and mating oppor- tunities in low-density populations.

Different processes may be responsible for site selection in different species. Even in 1 species, there may be several proximate and ultimate

Received 3 January 1989; revised 16 March 1989; accepted 16 March 1989.

causes of the behavior. The importance of these causes is likely to vary at different scales of hab- itat use (Morris, 1987). For example, the factors responsible for selection of a particular organ by a parasite may differ from the factors responsible for selection of specific sites within that organ. Explanations of site selection should therefore relate not only to proximate and ultimate causes but also to the scale at which the behavior is observed.

Adults of Echinococcus granulosus (Cestoda: Taeniidae) develop from ingested protoscoleces within the small intestine of domestic, feral, or wild canids. Previous studies have reported that worms are restricted to certain sites within the intestine of naturally or experimentally infected dogs. Counts of worms in segments of the small intestine usually have yielded small numbers in the duodenum and ileum and large numbers in the jejunum (Thompson and Eckert, 1983; Mac- pherson et al., 1985). Rausch (1985) proposed that site selection in Echinococcus functions to enhance cross-fertilization between worms. In support of this theory, he cited observations that adult worms occur in a more restricted area of the intestine in light than in heavy infections. Thompson and Lymbery (1988), however, pointed out that increased intraspecific compe- tition for space or nutrients at higher population densities may lead to increased niche breadth, and hence site extension, whatever the ultimate cause of site selection in the species.

Existing data on the dispersion of E. granu- losus within definitive hosts consist of counts of worms in segments of intestine. We describe a new approach that examines the spatial pattern

562

LYMBERY ET AL.-DISPERSION OF ECHINOCOCCUS 563

of individual worms attached to the intestinal wall of experimentally infected dogs. This may provide a more appropriate scale at which to examine the proximate and ultimate causes of site selection in the species.

MATERIALS AND METHODS Collection of protoscoleces

Protoscoleces were removed from hydatid cysts ob- tained from naturally infected sheep, cattle, and kan- garoos (Macropusfuliginosus). Protoscoleces from 1 or more cysts in the same organ of an individual host were regarded as 1 isolate. Nineteen isolates were ob- tained from 4 states and territories throughout Aus- tralia; 5 from Victoria, 5 from the Australian Capital Territory, 4 from New South Wales, and 5 from West- ern Australia. All isolates conformed morphologically to the Australian mainland sheep strain of E. granu- losus, as defined by Kumaratilake and Thompson (1984).

Infection of dogs Each isolate was used to infect 1 or more dogs; 26

dogs were infected from 19 isolates. Domestic cross- breeds of both sexes were used. At the time of infec- tion, they ranged in age from 3 to 12 mo. The history of previous infections was not known. On admission to the School of Veterinary Studies, Murdoch Univer- sity, all dogs were treated with praziquantel to remove adult tapeworms and their feces were examined regu- larly thereafter for the presence of tapeworm eggs. Each dog was infected orally with either 0.1, 0.2, 0.3, 0.4, 0.5, or 0.8 ml of protoscoleces, washed in phosphate- buffered saline, and packed in a gelatin capsule (0.1 ml is equivalent to approximately 40,000 protoscoleces [Thompson and Kumaratilake, 1985]). We obtained isolates at irregular intervals and infections took place over a 2-yr period. Despite this, all infections were carried out under standard conditions, immediately before feeding. After infection, all dogs were housed in similar conditions and maintained on a standard diet of tinned dog meat, dry biscuits, and water ad lib.

Autopsy and measurement of dispersion Thirty-five days after infection, dogs were killed by

an injection of pentobarbitone sodium, and the entire small intestine was removed. The intestine was opened longitudinally along the line of mesenteric attachment and divided transversely into thirds. A 50-mm length of tissue was removed from the anterior edge of each third, pinned to a piece of foam, and fixed in 10% formalin. The remainder of each third was incubated separately in a beaker of Hanks' balanced salt solution for 30 min at 37 C. The number of worms in each third of intestine was counted directly or estimated volumetrically.

Fixed intestinal sections, pinned to foam, were pho- tographed and enlarged to approximately 5 times their natural size. Photographs were overlain with a trans- parent grid of 5-mm x 5-mm contiguous quadrats and the number of worms in each quadrat scored. Worms were scored as belonging to a quadrat if more than half of their strobila was contained within it. Quadrats in

which worms were obscured, such as those along the edges of intestinal sections, were ignored in all analyses.

Analysis Data from quadrat samples were used to measure

the spatial pattern of worms attached to the intestinal wall. Several indices have been devised to describe, from quadrat counts or distance measures, the pattern or arrangement of individuals in a population and to test the departure of this pattern from randomness (see Patil and Stiteler, 1974; Southwood, 1978). All indices have limitations; we used the following 4 techniques because they differed in their assumptions and enabled the data to be analyzed in slightly different ways.

1) Negative binomial distributions were fitted to the data by the maximum likelihood method (Bliss and Fisher, 1953) and goodness of fit tested by x2. Aggre- gated patterns often follow the negative binomial dis- tribution and the exponent k is used as an index of aggregation; the smaller the value of k, the greater the extent of aggregation (Southwood, 1978). The mean size (number of individuals) of a clump in a negative binomial distribution was calculated as X = (x/2k)V (Arbous and Kerrich, 1951), where x is the mean num- ber of worms per quadrat and V is a x2 value at the 0.5 probability level with 2k degrees of freedom.

2) Morisita's test statistic (Morisita, 1959) was cal- q

culated as 16 = q[~ ni(ni - 1)/N(N - 1)], where q is

the number of quadrats, ni is the number of individuals in the ith quadrat, and N is the total number of indi- viduals. For a random pattern 16 = 1, values of 1b > 1 imply an aggregated pattern and values of 1b < 1 imply a regular pattern. The probability of observed departures from randomness can be obtained from Fo = [I6(N - 1) + q - N]/q - 1, with q - 1 and oo degrees of freedom.

3) Moran's coefficient of autocorrelation (Moran, q q

1950) was calculated as I = [q 2; wijzizj)]/ i=l j=l

q

[So 2 z2], where q is the total number of quadrats, wij i=O

is the weight of connection between the ith and jth quadrats, zi = ni - fi (ni is the number of individuals

q q

in the ith quadrat), and So = wij. Values of wij i=- j=l

were set to 1 for pairs of quadrats with adjacent edges, 0.5 for pairs with adjacent vertices, and 0 for all other pairs. I varies from - 1 to + 1, with the expected value in the absence of autocorrelation approaching 0 for large sample sizes; values of I greater than expected imply positive autocorrelation (adjacent quadrats with similar numbers of individuals), values less than ex- pected imply negative autocorrelation (adjacent quad- rats with dissimilar numbers of individuals). Standard deviates were calculated to measure the significance of departure from expectation (Cliff and Ord, 1981).

4) Taylor's power law (Taylor, 1961), s2 = acb or log s2 = log a + b log x, where x is the mean number of individuals per quadrat (density) and s2 is the vari- ance, was used to describe the relationship between dispersion and density over all samples. The slope of the regression (b) indicates the rate of increase of vari-

564 THE JOURNAL OF PARASITOLOGY, VOL. 75, NO. 4, AUGUST 1989

ance as mean density increases and is considered to be a measure of the density dependence of aggregation (Taylor et al., 1978, but see Downing, 1986). Ifb = 1, aggregation does not change with density, if b > 1 the population becomes more aggregated at higher densi- ties and if b < 1, the population becomes more regular at higher densities. Because both variance and mean density are subject to error, their relationship should be calculated as a functional, geometric mean regres- sion. In practice, however, this more accurate approach is seldom required (Southwood, 1978), and we have used predictive least squares regression analysis.

These 4 measures were obtained for a range of quad- rat sizes by combining adjacent quadrats and recal- culating number of individuals per quadrat. This pro- duced a series of 8 data sets per sample for fitting the negative binomial, Morisita's index and Taylor's pow- er curve, and 5 data sets/sample for spatial autocor- relation. Because quadrats along the edges of intestinal sections were not used, analyses did not account for 3- dimensional structure. This will introduce errors into all measures when using data from combined quadrats.

RESULTS

Recovery of adult worms

Adult worms were recovered from 23 infected dogs (88%). There was a significant linear rela- tionship between the logarithm of the number of adult worms recovered and the volume of pro- toscoleces ingested (log Y = 2.46 + 1.91 X, P (f = 0) < 0.05), although it accounted for only a small proportion of the variance in recovery (r2 = 0.18).

Dispersion of worms between intestinal segments

The number of adult worms in each third of the small intestine was scored in 19 infected dogs. In all cases, there were significant differences in the number of worms between segments (x2, P < 0.001). The mean proportion of worms in each segment was 0.69 in the first (most anterior), 0.27 in the second, and 0.03 in the third. Dispersion differed between dogs (heterogeneity x2 = 31,007, P < 0.001), with 15 dogs having the majority of worms in the first segment and 4 dogs having most worms in the second segment. Differences in dispersion did not appear to be related to sex or age of the hosts. There was no significant re- lationship between the total number of worms recovered and the proportion in each intestinal segment (Spearman's rank correlation: for the first segment rs = 0.13, P > 0.50; for the second rs = 0.18, P > 0.20; for the third rs = 0.25, P >

0.20).

Dispersion of worms within intestinal segments

The spatial pattern of worms was measured in 24 sections of intestine from 16 dogs. Fourteen

sections came from the first segment of intestine and 10 from the second. All indices of dispersion indicated an aggregated pattern at the smallest quadrat size. Morisita's Ib was significantly great- er than 1 in all samples. Moran's I was signifi- cantly greater than expected in 20 samples and showed the same trend in the other 4 samples. The negative binomial provided an adequate fit to the data from 22 samples. There was no evi- dence of difference in the degree of aggregation between samples from the first and second seg- ments of intestine, or between dogs of different sex or age.

In general, 16 decreased and k of the negative binomial increased as quadrat size increased. Morisita (1959) calculated the relationship be- tween 16 and quadrat size for various theoretical spatial patterns (see inset, Fig. 1). Plots from our data were of 2 types. Eighteen samples were con- sistent with Morisita's pattern for an aggregated dispersion with random intraclump spacing, whereas 6 samples conformed to the pattern for an aggregated dispersion with regular intraclump spacing. Plots for representative samples of each type are shown in Figure 1.

We used changes in spatial autocorrelation with quadrat size to estimate the size and area covered by a clump. Plots of I against quadrat size were of 3 types. In the most common pattern, shared by 12 samples, I remained constant or increased slightly over the first 3 quadrat sizes, then fell rapidly toward expected values. In 7 samples, I increased gradually and in the remaining 5 sam- ples decreased gradually, over all quadrat sizes. Plots for representative samples are shown in

Random

Regular

_ Aggregated

l .

. ..........................

fx~ Aggregated I

.reg.. 0 10 20 30 40 50 60 70

Area (mm 2)

FIGURE 1. Plot of Morisita's index (1I) against quadrat size (expressed as intestinal area, in mm2, after correction for photographic enlargement) for 2 samples of Echinococcus granulosus from the intestine of dogs. Open symbols indicate values significantly greater than 1. Inset: relationships between 16 and quadrat size for theoretical spatial patterns; random, regular, aggregat- ed with intraclump dispersion random, aggregated with intraclump dispersion regular.

0 .8 ....................... .......... . .. ............

0.8

X

_0

st

it

LYMBERY ET AL.-DISPERSION OF ECHINOCOCCUS 565

Figure 2. We took the quadrat size at which the last significant value of I occurred as a measure of the area covered by a clump. After correction for photographic enlargement, this ranged from 1 mm2 to 25 mm2, with a mean of 12.4 mm2 (SE = 1.8 mm2). The negative binomial distribution provided an adequate fit for the data from 22 samples at these quadrat sizes and the mean number of worms per clump, calculated by the method of Arbous and Kerrich (1951), ranged from 0.1 to 21.9, with an overall mean of 4.5 (SE = 1.2).

The sections of intestinal tissue that were used to measure spatial pattern were approximately 2,500 mm2 in area. With an average area/clump of 12.4 mm2, there were often several clumps in each section. These clumps could themselves ex- hibit a spatial pattern. It is difficult to obtain quantitative measures of such second-order pat- terns, but examination of original photographs and density maps computed from quadrat counts suggested a random dispersion of clumps in sec- tions of intestine (Fig. 3).

To determine the relationship between aggre- gation and density, we plotted values for our ag- gregation indices against mean number of indi- viduals per quadrat for each sample. Although Ib was negatively related to mean density, Iwao (1968) considered this an inherent property of the index. Moran's I was positively related to mean density; Figure 4 shows this relationship for the smallest quadrat size. This increase in degree of aggregation with increasing density was confirmed by Taylor's power law. Log variance was significantly related to log mean density at all quadrat sizes, with a positive intercept indi- cating aggregation. The slope, b, was significantly greater than 1 at all quadrat sizes except the smallest (Fig. 5).

DISCUSSION

Adult E. granulosus exhibited spatial pattern on 2 scales of habitat use. On a coarse scale, worms were found most frequently in the ante- rior third of the small intestine of infected dogs; we will refer to this as site restriction. On a finer scale, clumps of worms were found attached to the wall throughout the anterior two-thirds of the intestine, a phenomenon we refer to as ag- gregation. Site restriction and aggregation oc- curred in all dogs, regardless of sex or age. Site restriction may have been a consequence of the probability of attachment of evaginated proto- scoleces as they travelled along the intestine. This

0.2-

o 0.1-

0.0- -

-0.1 . , . 0 3 6 9 12 15 18

Area (mm 2 ) FIGURE 2. Plot of Moran's coefficient of autocor-

relation (I) against quadrat size (expressed as intestinal area, in mm2, after correction for photographic en- largement) for 3 samples of Echinococcus granulosus from the intestine of dogs. Open symbols indicate val- ues significantly greater than expected.

seems unlikely, however, because both the pres- ent study and that of Macpherson et al. (1985) found major differences in dispersion of worms between intestinal segments in different dogs un- der identical conditions of infection. Similarly, it is possible that aggregation resulted from clumping of protoscoleces within hydatid cysts, but the protoscoleces should have been removed from their brood capsules and thoroughly dis- persed, both by washing prior to infection and by the action of pepsin in the stomach (Smyth, 1969). Presumably then, site restriction and ag- gregation resulted from the behavior of devel- oping worms. We will attempt to infer the prox- imate and ultimate causes of this behavior, proceeding from finer to coarser scales of habitat use.

The proximate cause of aggregative behavior may be either attraction between individuals or attraction to patchy microenvironmental factors. Clumps of about 4-5 individuals occurred throughout a substantial portion of the small in- testine. They occupied an area of about 12 mm2 and there was no observable second-order pat- tern to their occurrence within sections of intes- tine. These data are more consistent with at- traction between individuals than with attraction to environmental factors, especially if structural and physicochemical gradients in the intestine of vertebrates are essentially linear, as proposed by Read (1971), Crompton (1973), and Mettrick and Podesta (1974).

The ultimate cause of aggregative behavior in E. granulosus is more difficult to infer. At least 5 biological functions of such behavior have been suggested for other organisms; choice of an op- timal habitat (Price, 1984), avoidance of pre-

566 THE JOURNAL OF PARASITOLOGY, VOL. 75, NO. 4, AUGUST 1989

a

I,

No data

No worms

1 worm

2 worms

3 worms

4 worms

D lI

LI *

LYMBERY ET AL.-DISPERSION OF ECHINOCOCCUS 567

22 Li0 0

1 0- 0

0

00 0

o

.01 .01 .1 1 10

Mean Density

FIGURE 4. Relationship between Moran's I and mean density (number of worms per quadrat) at the smallest quadrat size for 24 samples of Echinococcus granulosus from the intestine of dogs. Regression line described by log Y = -0.73 + 0.40 log X (r2 = 0.41, P ( = 0) < 0.001).

dation (Bertram, 1978), avoidance of interspe- cific competition (Holmes, 1973), enhancement of mating opportunities (Rohde, 1982), and im- provement of environmental quality (Way and Cammell, 1970). Three of these can be ruled out for E. granulosus. Firstly, because of the small size of clumps and their occurrence throughout the small intestine, aggregative behavior is un- likely to result in optimal habitat use. Secondly, we are not aware of any predators (or parasites) that could have been responsible for the evolu- tion of the behavior. Thirdly, the occurrence of aggregations in the absence of other species in- dicates that aggregative behavior is not primarily an interactive response to interspecific compe- tition. It may be regarded as a selective response if the behavior minimizes niche overlap between E. granulosus and other species of intestinal hel- minths. This is unlikely, however, if the proxi- mate cause of the behavior is attraction between individuals, because the broad fundamental niche of E. granulosus means that the presence of con- specifics will not provide a reliable cue to the absence of interspecific competitors. Kuno (1988) showed that increasing the patchiness of a species distribution relaxes interspecific competition, even when 2 species share the same niche. In his model, however, this occurs because of a low- ering of equilibrium density through increased intraspecific competition within patches. Aggre- gative behavior could not have evolved through

0.8-

1.6 -

b 1.4-

0.8 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70

Area (mm 2

)

FIGURE 5. Slope (b) of the regression of log variance on log mean density at different quadrat sizes (ex- pressed as intestinal area, in mm2, after correction for photographic enlargement). b is significantly greater than 0 at all quadrat sizes and significantly greater than 1 at all except the first.

individual selection to enhance intraspecific competition, although that may well be an effect (sensu Williams, 1966) of the trait.

The observed spatial pattern and likely prox- imate cause of aggregative behavior in E. gran- ulosus are consistent with both of the remaining functions. Enhancement of mating opportunities requires that cross-fertilization occurs between worms. This has never been observed directly in E. granulosus, although self-fertilization (or at least self-insemination) has been reported in a number of studies (Kumaratilake et al., 1986). Data on the genetic variation of E. granulosus in Australia suggest that both cross- and self- fertilization may occur in natural populations (Lymbery and Thompson, 1988). Environmen- tal improvement could take the form of changes in gut morphology near groups of worms, pro- viding easier attachment, or of changes in nu- trient concentrations through increased flow in the intestinal lumen or increased breakdown of mucosal cells. However, we are not aware of any evidence that such changes occur in hosts in- fected with E. granulosus.

The degree of aggregation increased with in- creasing density of worms in the intestine. Taylor and Taylor (1977) interpreted this phenomenon, which has been observed in a wide range of free- living species, as evidence that the function of aggregative behavior is to maximize environ- mental quality. Anderson et al. (1982) and

FIGURE 3. a. Photograph of a section from the first third of the small intestine from a dog, 35 days after infection with 0.5 ml of protoscoleces of Echinococcus granulosus. Scale bar = 10 mm. b. Density map, computed from the photograph by counting the number of worms in 1-mm2 quadrats.

9. CI

568 THE JOURNAL OF PARASITOLOGY, VOL. 75, NO. 4, AUGUST 1989

Downing (1986), however, pointed out that den- sity dependence of aggregation may arise from the stochastic interplay of demographic events and environmental heterogeneity, and it need not imply an evolved behavioral response.

Although the observed relationship between the degree of aggregation and density cannot be used to support either of the likely ultimate ex- planations of aggregative behavior in E. granu- losus, these explanations do make different pre- dictions about the degree of aggregation in response to development of worms and genetic heterogeneity of the initial infection. If aggre- gative behavior increased around the time of sex- ual maturity, this would be consistent with the enhancement of mating opportunities but not with improvement of environmental quality. Al- though there is no evidence for extensive migra- tions in the intestine, worms are able to move between adjacent villi as soon as their hooks and suckers can be used for attachment (Smyth et al., 1969). Sexual maturity is reached about 28 days after infection with the sheep strain of E. gran- ulosus (Smyth et al., 1969) and there is evidence that movement declines soon afterwards (Thompson et al., 1979). We have attempted re- cently to measure spatial pattern in dogs autop- sied 20 and 25 days after infection, but worms were too small to be visible consistently above the villi of the intestine. If the function of ag- gregative behavior is to enhance mating between worms, and if cross-fertilization serves to pro- mote outbreeding, then the degree of aggregation may be expected to be greater in genetically het- erogeneous populations than in populations of genetically identical individuals derived from a single cyst (assuming that worms are capable of distinguishing different genotypes). This would not be predicted if the function of aggregative behavior is to improve environmental quality. We found no difference in the degree of aggre- gation between infections derived from a single cyst and those derived from a number of cysts from the same host, but we have no measure of genotypic diversity in the latter case.

As well as forming clumps on the wall of the small intestine of infected dogs, adult worms oc- curred disproportionately along the length of the intestine. This site restriction may be a by-prod- uct of aggregative behavior on a finer scale. Al- ternatively, the proximate factors responsible for site restriction may operate independently of those causing aggregative behavior. The best evi- dence that different factors are responsible for

spatial pattern at different scales of habitat use comes from studies that have reported a decrease in site restriction at higher densities (Sweatman and Williams, 1963; Macpherson et al., 1985; Rausch, 1985; Gemmell et al., 1986). If this oc- curs, it must be through processes operating in- dependently of those on a finer scale because we found that the degree of aggregation increased with density. An analysis of previous studies, however, shows that there is no published evi- dence to support a relationship between density and site restriction for Echinococcus. Most re- ports are anecdotal. Macpherson et al. (1985: table 3) provided data (pooled from a number of experimentally infected dogs) on the percent- ages of worms in 4 regions of the small intestine. Although pooling may have obscured trends in individual dogs, there is no significant relation- ship between total number of worms recovered and percentage in each region of the intestine (Spearman's rank correlation: for the first region rs = 0.50, P > 0.05; for the second r, = 0.30, P > 0.10; for the third r, = 0.43, P > 0.05; for the fourth r5 = 0.37, P > 0.10). In the present study, we also found no relationship between number of worms in the intestine and dispersion between intestinal segments.

Rausch (1985) suggested that the restriction of adult Echinococcus to certain regions of the small intestine of dogs functions to enhance cross-fer- tilization. However, until it can be shown that site restriction is not simply a consequence of aggregative behavior, a search for ultimate ex- planations of the phenomenon seems premature. We believe that it would be more rewarding for further analyses of the dispersion of E. granu- losus in the intestine of dogs to concentrate on the finer scale of habitat use.

ACKNOWLEDGMENTS

We thank Jon Dunsmore for comments on the manuscript and Aileen Elliot for invaluable tech- nical assistance. Hydatid material was kindly provided by B. Dennis, B. Huisman, D. Jenkins, R. Lyford, P. McGregor, D. Rennell, and R. Stanton. This work was supported by a grant from the Australian Research Grants Scheme and by National Research and Australian Research Council Fellowships to Alan Lymbery.

LITERATURE CITED

ANDERSON, R. M., D. M. GORDON, M. J. CRAWLEY, AND M. P. HASSELL. 1982. Variability in the

LYMBERY ET AL.-DISPERSION OF ECHINOCOCCUS 569

abundance of animal and plant species. Nature 296: 245-248.

ARBOUS, A. G., AND J. E. KERRICH. 1951. Accident statistics and the concept of accident-proneness. Biometrics 7: 340-432.

BERTRAM, B. C. R. 1978. Living in groups: Predators and prey. In Behavioural ecology, J. R. Krebs and N. B. Davies (eds.). Blackwell Scientific Publica- tions, Oxford, p. 64-96.

BLISS, C. I., AND R. A. FISHER. 1953. Fitting the neg- ative binomial distribution to biological data and note on the efficient fitting of the negative bino- mial. Biometrics 9: 176-200.

CLIFF, A. D., AND J. K. ORD. 1981. Spatial processes. Pion Limited, London, 266 p.

CROMPTON, D. W. T. 1973. The sites occupied by some parasitic helminths in the alimentary tract of vertebrates. Biological Reviews 48: 27-83.

DOWNING, J. A. 1986. Spatial heterogeneity: Evolved behaviour or mathematical artefact? Nature 323: 255-257.

GEMMELL, M. A., J. R. LAWSON, AND M. G. ROBERTS. 1986. Population dynamics in echinococcosis and cysticercosis: Biological parameters of Echinococ- cus granulosus in dogs and sheep. Parasitology 92: 599-620.

HOLMES, J. C. 1973. Site selection by parasitic hel- minths: Interspecific interactions, site segregation, and their importance to the development of hel- minth communities. Canadian Journal of Zoology 51: 333-347.

IWAO, S. 1968. A new regression method for analyz- ing the aggregation pattern of animal populations. Researches on Population Ecology 10: 1-20.

KUMARATILAKE, L. M., AND R. C. A. THOMPSON. 1984. Morphological characterisation of Australian strains of Echinococcus granulosus. International Journal for Parasitology 14: 467-477.

, J. ECKERT, AND A. D'ALESSANDRO. 1986. Sperm transfer in Echinococcus (Cestoda: Taeniidae). Zeitschrift fur Parsitenkunde 72: 265- 269.

KUNO, E. 1988. Aggregation pattern of individuals and the outcomes of competition within and be- tween species: Differential equation models. Re- searches on Population Ecology 30: 69-82.

LYMBERY, A. J., AND R. C. A. THOMPSON. 1988. Elec- trophoretic analysis of genetic variation in Echino- coccus granulosus from domestic hosts in Austra- lia. International Journal for Parasitology 18: 803- 811.

MACPHERSON, C. N. L., C. M. FRENCH, P. STEVENSON, L. KARSTAD, AND J. H. ARUNDEL. 1985. Hydatid disease in the Turkana District of Kenya, IV. The prevalence of Echinococcus granulosus infections in dogs, and observations on the role of the dog in the lifestyle of the Turkana. Annals of Tropical Medicine and Parasitology 79: 51-61.

METTRICK, D. F., AND R. B. PODESTA. 1974. Ecolog- ical and physiological aspects of helminth-host interactions in the mammalian gastrointestinal ca- nal. Advances in Parasitology 12: 183-278.

MORAN, P. A. P. 1950. Notes on continuous sto- chastic phenomena. Biometrika 37: 17-23.

MORISITA, M. 1959. Measuring of the dispersion of

individuals and analysis of the distributional pat- terns. Memoirs of the Faculty of Science, Kyushu University, Series E (Biology) 2: 215-235.

MORRIS, D. W. 1987. Ecological scale and habitat use. Ecology 68: 362-369.

PATIL, G. P., AND W. M. STITELER. 1974. Concepts of aggregation and their quantification: A critical review with some new results and applications. Researches on Population Ecology 15: 238-254.

PRICE, P. W. 1984. Communities of specialists: Va- cant niches in ecological and evolutionary time. In Ecological communities: Conceptual issues and the evidence, D. R. Strong, Jr., D. Simberloff, L. G. Abele, and A. B. Thistle (eds.). Princeton Uni- versity Press, Princeton, New Jersey, p. 510-523.

RAUSCH, R. L. 1985. Parasitology: Retrospect and prospect. Journal of Parasitology 71: 139-151.

READ, C. P. 1971. The microcosm of intestinal hel- minths. In Ecology and physiology of parasites, A. M. Fallis (ed.). University of Toronto Press, Toronto, p. 188-200.

ROHDE, K. 1979. A critical evaluation of intrinsic and extrinsic factors responsible for niche restric- tion in parasites. American Naturalist 114: 648- 671.

1982. Ecology of marine parasites. Univer- sity of Queensland Press, St. Lucia, Queensland, 245 p.

, AND R. P. HOBBS. 1986. Species segregation: Competition or reinforcement of reproductive barriers? In Parasite lives, C. Cremin, C. Dobson, and D. E. Moorhouse (eds.). University of Queens- land Press, St. Lucia, Queensland, p. 189-199.

SMYTH, J. D. 1969. The physiology of cestodes. Oli- ver and Boyd, Edinburgh, 279 p.

, M. GEMMELL, AND M. M. SMYTH. 1969. Es- tablishment of Echinococcus granulosus in the in- testine of normal and vaccinated dogs. Indian Journal of Helminthology, H. D. Srivastava Com- memoration Volume, p. 167-178.

SOUTHWOOD, T. R. E. 1978. Ecological methods. Chapman and Hall, London, 524 p.

SUKHDEO, M. V. K., AND D. F. METTRICK. 1987. Par- asite behaviour: Understanding platyhelminth re- sponses. Advances in Parasitology 26: 73-144.

SWEATMAN, G. K., AND R. J. WILLIAMS. 1963. Com- parative studies on the biology and morphology of Echinococcus granulosus from domestic live- stock, moose and reindeer. Parasitology 53: 339- 390.

TAYLOR, L. R. 1961. Aggregation, variance and the mean. Nature 189: 732-735.

, AND R. A. J. TAYLOR. 1977. Aggregation, migration and population mechanics. Nature 265: 415-421.

, I. P. WOIWOD, AND J. N. PERRY. 1978. The density-dependence of spatial behaviour and the rarity of randomness. Journal of Animal Ecology 47: 383-406.

THOMPSON, R. C. A., J. D. DUNSMORE, AND A. R. HAYTON. 1979. Echinococcus granulosus: Secre- tory activity of the rostellum of the adult cestode in situ in the dog. Experimental Parasitology 48: 144-163.

, AND J. ECKERT. 1983. Observations on

570 THE JOURNAL OF PARASITOLOGY, VOL. 75, NO. 4, AUGUST 1989

Echinococcus multilocularis in the definitive host. Zeitschrift fuir Parasitenkunde 69: 335-345.

, AND L. M. KUMARATILAKE. 1985. Compar- ative development of Australian strains of Echi- nococcus granulosus in dingoes (Canis familiaris dingo) and domestic dogs (C. f familiaris), with further evidence for the origin of the Australian sylvatic strain. International Journal for Parasi- tology 15: 535-542.

, AND A. J. LYMBERY. 1988. The nature, extent and significance of variation within the genus

Echinococcus. Advances in Parasitology 27: 209- 258.

WAY, M. J., AND M. CAMMELL. 1970. Aggregation behaviour in relation to food utilization by aphids. In Animal populations in relation to their food resources, A. Watson (ed.). Blackwell Scientific Publications, Oxford, p. 229-247.

WILLIAMS, G. C. 1966. Adaptation and natural se- lection. Princeton University Press, Princeton, New Jersy, 307 p.